ORIGINAL ARTICLE

Phylogenetics and predictive distribution modeling provideinsights into the geographic divergence of Eriosyce subgen.Neoporteria (Cactaceae)

Pablo C. Guerrero • Mary T. K. Arroyo •

Ramiro O. Bustamante • Milen Duarte •

Thomas K. Hagemann • Helmut E. Walter

Received: 25 June 2010 / Accepted: 22 July 2011 / Published online: 13 August 2011

� Springer-Verlag 2011

Abstract The classification of Eriosyce subgenus Neo-

porteria (‘‘subsection’’ in the sense of Kattermann) and the

role of allopatry/sympatry in the diversification of the

group were studied by use of cladistic and predictive dis-

tribution modeling methods. We reconstructed the phylo-

genetic relationships of subgenus Neoporteria by analyzing

38 morphological characters and DNA sequences from two

chloroplast regions of 21 taxa from the Chilean subsections

of Eriosyce using a Bayesian and maximum likelihood

phylogenetic framework. Also, we attempted to find out if

the divergence between the sister taxa in the Neoporteria

group had been caused by allopatric or sympatric mecha-

nisms. The morphology-based analysis placed E. chilensis

basal within the Neoporteria clade and suggested a further

broadening of the group by including E. taltalensis var.

taltalensis, formerly considered a member of subsection

Horridocactus. However, the combined DNA data placed

E. sociabilis and E. taltalensis var. taltalensis within the

Horridocactus clade, and placed E. chilensis with E. sub-

gibbosa var. litoralis. The broad concept of E. subgibbosa

sensu Kattermann (comprising seven infraspecific taxa),

was rejected by our combined molecular results. Finally,

our results corroborated changes in subsection Neoporteria

proposed by various authors and suggested further modi-

fications within Neoporteria. The analyses of the degree of

geographic overlap of the predicted distributions indicated

null overlap between the sister taxa, and one probable

hybrid origin of E. chilensis, indicating that evolutionary

divergence is mainly caused by an allopatric process

associated with climatic tolerance.

Keywords Chile � Morphology � Diversification �Ecological niche modeling � Vicariance � Speciation

Introduction

Understanding the processes that lead to the origin of new

taxa is a fundamental objective of evolutionary biology

(Darwin 1859; Futuyma 2005). In this sense, the essential

role of geographic isolation driving divergence of taxa had

become widely accepted by the mid-20th century (Mayr

1959). Recently, new support for sympatric speciation

mediated by disruptive phenotypic selection exerted by

pollinators challenged geographic isolation as the single

force driving the formation of new lineages (Schemske and

Bradshaw 1999; Turelli et al. 2001; Fitzpatrick and Turelli

2006; Savolainen et al. 2006). Moreover, the chance of

sympatric speciation in organisms with a low mobility (for

example plants) is higher, and it also increases the possi-

bility of making accurate inferences of the geography of

speciation (Losos and Glor 2003). For example, if two

sister plant species (both with limited seed dispersal) are

distributed allopatrically, it is plausible to conclude that

their divergence mechanism must be associated with geo-

graphic isolation (Barraclough and Vogler 2000; Losos and

Electronic supplementary material The online version of thisarticle (doi:10.1007/s00606-011-0512-5) contains supplementarymaterial, which is available to authorized users.

P. C. Guerrero (&) � M. T. K. Arroyo �R. O. Bustamante � M. Duarte � T. K. Hagemann

Departamento de Ciencias Ecologicas, Instituto de Ecologıa y

Biodiversidad, Facultad de Ciencias, Universidad de Chile,

Casilla 653, 7800024 Santiago, Chile

e-mail: pablo.c.guerrero@gmail.com

H. E. Walter

The EXSIS Project: Ex-situ & in-situ Cactaceae Conservation

Project, Santiago, Chile

e-mail: helmut_walter1@hotmail.com

123

Plant Syst Evol (2011) 297:113–128

DOI 10.1007/s00606-011-0512-5

Glor 2003), whereas growing in sympatry ecological

mechanisms, for example pollinator guilds, might account

for their divergence (Schemske and Bradshaw 1999). A

key issue highlighted by speciation biologists is the rele-

vance of studying the process of divergence in a lineage-

based concept (Wiens 2004a), linking methodological tools

that enable understanding of the mechanism(s) underlying

the evolutionary and the ecological processes that split an

ancestral species into new lineages (Wiens 2004b). These

views may be addressed by use of phylogenetic tools and

predictive distribution modeling methods accounting for

geographic dimensions of evolution (Graham et al. 2004;

Kozak and Wiens 2006; Posadas et al. 2006; Katinas and

Crisci 2008; Evans et al. 2009). Moreover, predictive dis-

tribution modeling has also demonstrated its utility by

helping to solve problems arising from early stages of

divergence or postspeciation, distinguishing cryptic species

linages when morphological differences are subtle, over-

lapping, or not yet fixed (Raxworthy et al. 2007), because it

provides evidence for geographic isolation between popu-

lations, and hence provides evidence for these populations

being viewed as separate evolving lineages when gene flow

is considered unlikely for the intervening unsuitable region

(Graham et al. 2004; Wiens and Graham 2005; Raxworthy

et al. 2007; Evans et al. 2009).

Eriosyce sensu lato (=sensu Kattermann) is a large

South American genus of subglobular, globular or elon-

gated cacti belonging to tribe Notocacteae and comprising

some 30 species in its current circumscription (Katter-

mann 1994; Anderson 2001; Hoffmann and Walter 2004;

Hunt et al. 2006). Most of its members occur in Chile

between latitudes 18� and 36� S, but also in southern Peru

between 14� and 18� S and in western Argentina from 24�to 37� S. In 1994, Kattermann proposed this broad con-

cept of Eriosyce, amplifying the genus by including the

former genera Horridocactus, Islaya, Neoporteria, Pyr-

rhocactus, and Thelocephala. Eriosyce sensu lato is

divided into section Eriosyce, comprising the subsections

Eriosyce sensu stricto (endemic to Chile), Pyrrhocactus

(endemic to Argentina), and Islaya (extinct in the wild in

northern Chile but frequent in southern Peru) and section

Neoporteria (all taxa endemic to Chile) comprising sub-

sections Neoporteria, Horridocactus, and Chileosyce.

While this broad concept of the genus Eriosyce is widely

used (but see Zuloaga et al. 2008), Kattermann’s infra-

generic and infraspecific classification of Eriosyce has

been challenged by several authors (Nyffeler and Eggli

1997, 2010; Hunt 2003; Ferryman 2003; Hoffmann and

Walter 2004; Hunt et al. 2006; Walter and Machler 2006;

Walter 2008).

Subsection Neoporteria is one of the most conspicuous

groups within Eriosyce sensu lato and is mainly distributed

in coastal areas, but sometimes also in transversal valleys

with coastal influence between the latitudes of 27� and 36�S. With the exception of E. chilensis, its narrow funnel-

form to tubular fuchsia-colored flowers are hummingbird-

pollinated and their stems become elongated with age—

sometimes up to 1 m in length. Neoporteria occupies areas

with elevated oceanic influence in the southerly Atacama

Desert and the Mediterranean zones of Chile (Luebert and

Pliscoff 2006). Some taxonomic proposals were based on

incomplete taxa sampling and they used a limited number

of floral characters that made it difficult to define infra-

generic and interspecific limits, and several phylogenetic

aspects remained unsolved.

In this study we examine the specific and infraspecific

classification of subsection Neoporteria and test the

hypothesis of its monophyly against members from the

other Chilean subsections of Eriosyce sensu lato. We

attempted to determine the infrageneric position of

E. chilensis and E. taltalensis within Eriosyce sensu lato by

using molecular and morphological data. Because none of

the previous taxonomic or phylogenetic proposals have

been spatially evaluated, thus providing little evidence for

geographic isolation between sister taxa or the process of

ongoing evolutionary divergence, we attempted to eluci-

date the role of allopatry/sympatry in the divergence found

within subsection Neoporteria by comparing the predicted

distribution overlap between sister taxa using predictive

distribution modeling.

Materials and methods

Plant material

The morphological dataset was obtained by measuring at

least 20 individuals in the field in Neoporteria sensu stricto,

the E. taltalensis complex, and the specimens detailed in

Table 1. We measured 10 individuals of each species of the

other Chilean subsections of Eriosyce. In the case of

E. islayensis (a taxon extinct in the wild in Chile) we used

material grown from documented habitat seed (FK

895 = Fred Kattermann) and data taken from literature;

data for C. brevistylus were taken from cultivated

ex-habitat specimens. Most of the characters of the epi-

dermis, hypodermis, cortex, and pith were adopted from

Nyffeler and Eggli (1997).

Taxon sampling

The sampling of the Eriosyce taxa used in this study was

based on the classification in Kattermann (1994, 2001)

because it is widely used and enabled us to make com-

parisons with the first phylogenetic hypothesis proposed for

Eriosyce sensu lato. Twelve distinct operational taxonomic

114 P. C. Guerrero et al.

123

Table 1 List of taxa, their infrageneneric categories used in Kattermann (1994), their acronyms used in the cladistic and distribution modeling

analyses, their latitudinal distribution in Chile, and the plant material used in this study

Species Acronyms Latitudinal

distribution

Collection data

Outgroups

Corryocactus brevistylus (K. Schumann)

Britton and Rose

C_brevistylus 18�–20� S HW 74 (CONC) [JF975686, JF975703], Region de Tarapaca, west of

Chusmiza. F. Ritter 122 a (SGO), Region de Tarapaca, Camina

Neowerdermannia chilensis Backeb. Neo_chilensis 17�–18� S HW 860 (CONC) [–, JF975704], Region de Tarapaca, Socoroma 3200 m.

F. Ritter 199 (SGO), Region de Tarapaca, Chapiquina.

Subsection Eriosyce

E. aurata (Pfeiff.) Backeb. E_aurata 27�–33� S HW 17 (CONC) [JF975687, JF975706], Region de Coquimbo, Fray Jorge.

Subsection Horridocactus

E. curvispina var. curvispina (Bertero ex

Colla) Katt.

E_curvispina_cur 30�–35� S HW 23 (CONC) [JF975689, JF975708], Region Metropolitana, El Volcan.

E. odieri subsp. odieri (Lem. ex Salm-

Dyck) Katt.

E_odieri_odi 27� S HW 119 (CONC) [–, JF975711], Region de Atacama, Puerto Viejo.

Eriosyce taltalensis (Hutchison) Katt.

subsp. paucicostata (F.Ritter) Katt.

E_tal_pau 24�–25� S HW 237 a (CONC) [–, JF975705], Region de Antofagasta, north of Paposo.

HW 94, Region de Antofagasta, east of Paposo; HW 422, Region de

Antofagasta, east of Taltal. F. Ritter (SGO), Region de Antofagasta, 22 km

north of Paposo.

E. taltalensis (Hutchison) Katt. E_tal_tal 25�–27� S FK 1061 (SGO), Region de Antofagasta, Esmeralda.

Eggli 2922 (CONC, SGO), Region de Antofagasta, near Taltal branching of the

Quebrada Los Changos and Quebrada Los Andes.

HW 253 (CONC) [JF975701, JF975724], Region de Antofagasta, Esmeralda.

HW 754, Region de Antofagasta, north of Paposo.

HW 400, Region de Antofagasta, Barquito

E. taltalensis (Hutchison) Katt. subsp.pilispina (F.Ritter) Katt.

E_tal_pil 26� S HW 415, Region de Atacama, above Barquito, 500 m.

HW 545 (CONC) [–, JF975722], Region de Atacama, 5 km east of Obispito.

E. taltalensis (Hutchison) Katt. var.pygmaea (F.Ritter) Katt.

E_tal_pyg 26�–27� S HW 398 (CONC) [–, JF975723], Region de Atacama, Pan de Azucar. HW 103,

Region de Atacama, north of Chanaral. HW 227, Region de Atacama, Las

Conchillas. HW 379, Region de Atacama, Morro Copiapo.

Subsection Islaya

E. islayensis (C.F. Forster) Katt. E_islayensis 14�–18� S FK 895 (CONC) [JF975690, JF975709], Southern Peru, Arequipa.

Specimen grown from habitat seed.

Subsection Chileosyce

E. krausii (F. Ritter) Katt. E_krausii 26� S F. Ritter 502 (SGO), Region de Atacama, north of Caldera. HW 262 (CONC) [–

,JF975710], Region de Atacama, north of Flamenco. HW 265, Region de

Atacama, Porto Fino.

HW 736, Region de Atacama, Pan de Azucar.

Ingroup

Subsection Neoporteria

E. chilensis (Hildm. ex K. Schumann)

Katt.

E_chilensis 32� S FK 3 (SGO) (Neotype), Region de Valparaıso, Los Molles.

FK 192 (SGO), Region de Coquimbo, Pichidangui.

Eggli and Leuenberger 2931 (CONC), Region de Valparaıso, Los Molles.

Eggli and Leuenberger 2934 (CONC), Region de Coquimbo, Pichidangui.

HW 609 (CONC) [JF975688, JF975707], Region de Coquimbo, south of

Pichidangui.

E. senilis subsp. coimasensis (F. Ritter)

Katt.

E_senilis_coi 32� S FK 266 (SGO), Region de Valparaıso, Las Coimas.

HW 9 (CONC) [JF975691, JF975712], Region de Valparaıso, Las Coimas.

E. senilis subsp. senilis (Backeb.) Katt. E_senilis_sen 30�–32� S FK 29 (SGO), Region de Coquimbo, 43 km east of La Serena.

FK 418 (SGO), Region de Coquimbo, Cuncumen.

FK 422 (SGO) (Neotype), Region de Coquimbo, West of Coyton

(sic) = Coiron.

FK 423 (SGO), Region de Coquimbo, Quelen.

FK 452 (SGO), Region de Coquimbo, Huampulla.

FK 462 (SGO), Region de Coquimbo, El Tambo.

Eggli and Leuenberger 2553 (SGO), Region de Coquimbo, Cuesta El Espino.

HW 636 (CONC) [JF975692, JF975713], Region de Coquimbo, Huampulla.

Phylogenetics and predictive distribution modeling 115

123

Table 1 continued

Species Acronyms Latitudinal

distribution

Collection data

E. sociabilis (F. Ritter) Katt. E_sociabilis 27� S FK 147 (SGO) (Neotype), Region de Atacama, Totoral Bajo.

FK 803 (SGO), Region de Atacama, Morro Copiapo.

HW 279 (CONC) [JF975693, JF975714], Region de Atacama, south of Totoral

Bajo.

HW 387, Region de Atacama, north of Caldera.

HW 526, Region de Atacama, east of Obispito.

HW 646, Region de Atacama, north of Totoral Bajo.

E. subgibbosa (Haw.) Katt. E_subgibbosa_sub 32�–36� S FK 207 (SGO), Region de Coquimbo, 10 km north of Valparaıso.

Eggli and Leuenberger 2938 (CONC), Region de Coquimbo, Pichidangui.

Eggli, Leuenberger and Arroyo-Leuenberger 3089 (SGO), Region de

Coquimbo, Caleta Oscuro.

Eggli, Leuenberger and Arroyo-Leuenberger 3104 (CONC), Region de

Valparaıso, Quinteros.

Eggli and Leuenberger 3109 (CONC), Region de Valparaıso, El Quisco.

Eggli, Leuenberger and Arroyo-Leuenberger 3119 (CONC), Region del

Libertador General Bernardo O’Higgins, Punta de Lobos.

Eggli, Leuenberger and Arroyo-Leuenberger 3122 (CONC), Region del

Libertador General Bernardo O’Higgins, Bucalemu.

Eggli, Leuenberger and Arroyo-Leuenberger 3123 (CONC), Region del

Libertador General Bernardo O’Higgins, Bucalemu.

Eggli, Leuenberger and Arroyo-Leuenberger 3134 (CONC), Region del Maule,

Limpimavida.

K. Behn s.n. (CONC), Region de Coquimbo, Cerro La Cruz. K. Behn s.n.

(CONC), Region de Valparaıso, Concon.

HW 147 (CONC), Region de Valparaıso, south of Valparaıso.

HW 594, Region del Maule, south of Constitucion.

HW 645, Region de Coquimbo, Confluencia.

HW 214 [JF975698, JF975719], Region del Libertador General Bernardo

O’Higgins, Punta de Lobos.

E. sugibbosa var. castanea (F. Ritter)

Katt.

E_sugibbosa_cas 34�–35� S FK 202 (SGO), Region del Libertador General Bernardo O’Higgins, Santa

Cruz.

Eggli, Leuenberger and Arroyo-Leuenberger 3139 (CONC), Region del Maule,

18 km south of San Javier.

FK 204 (SGO), Region del Maule, Villa Prat.

HW 34 (CONC) [JF975694, JF975715], Region del Maule, Villa Prat.

HW s.n., Region del Libertador General Bernardo O’Higgins, Tuna.

HW 155, Region del Maule, Litu.

E. subgibbosa var. litoralis (F.Ritter)

Katt.

E_subgibbosa_lit 29�–31� S HW 47 (CONC), Region de Coquimbo, Totoralillo.

HW 65 Region de Coquimbo, south of Chungungo.

HW 619 [JF975696, JF975717] Region de Coquimbo, north of Los Vilos.

E. subgibbosa subsp. clavata (Sohrens

ex K. Schum.) Katt.

E_subgibbosa_cla 29�–30� S FK 27 (SGO) (Neotype), Region de Coquimbo, 30 km east of La Serena.

Eggli and Leuenberger 3071 (CONC), Region de Coquimbo, 19 km west of La

Serena.

Eggli and Leuenberger 3079 (CONC), Region de Coquimbo, Quebrada

Marquesa.

HW 46 (CONC) [JF975695, JF975716], Region de Coquimbo, Las Rojas.

HW 45 Region de Coquimbo, Cuesta San Antonio.

E. subgibbosa subsp. nigrihorrida(Backeb. ex A.W.Hill) Katt.

E_subgibbosa_nig 30�–31� S FK 24 (SGO), Region de Coquimbo, Tongoy.

FK 84 (SGO), Region de Coquimbo, Limarı river.

FK 481 (SGO), Region de Coquimbo, Punta Teniente.

FK 1095 (SGO), Region de Coquimbo, Limarı river.

HW 628 (CONC) [JF975697, JF975718], Region de Coquimbo, El Teniente.

HW 62, Region de Coquimbo, Cuesta las Cardas.

HW 730, Region de Coquimbo, Quebrada Seca.

116 P. C. Guerrero et al.

123

units from subsection Neoporteria at species and infra-

specific levels were chosen as an ingroup in all the phy-

logenetic analyses (recent taxonomic proposals of ingroups

and outgroups are available in the electronic supplementary

material, Appendix 1). Infraspecific taxa were included for

two reasons: first, because of the problem of polymorphic

character states within species (Nixon and Davis 1991),

and second, because earlier studies suggested that some

taxa previously classified at infraspecific levels might

represent distinct species (Nyffeler and Eggli 1997;

Walter 2008) or might belong to different groups (Walter

2008; Appendix 1, supplementary material). Several spe-

cies from the Chilean subsections of Eriosyce sensu lato

(subsection Eriosyce, subsection Islaya, subsection Hor-

ridocactus and subsection Chileosyce) were used as out-

groups. In the morphological analysis we included eight

outgroup species from Eriosyce sensu lato, four in the

trnL-trnF and nine in the rpl32-trnL data set. We also

included distantly related Corryocactus brevistylus (tribe

Phyllocacteae subtribe Corryocactinae) in all analyses,

and Neowendermannia chilensis in analyses that included

rpl32-trnL. Voucher information and a complete list of the

21 Eriosyce sensu lato taxa with their collection data and

the acronyms used in the analyses and their distribution

are shown in Table 1.

Morphological dataset

The dataset consisted of 38 discrete multistate morpho-

logical characters. We decided to include eco-physiologi-

cal characters because ex-habitat and seed-grown plants

cultivated in the northern hemisphere had the same char-

acters, suggesting a genetic background for those charac-

ters (Walter 2008). Thickness of epidermis and thickness of

hypodermal layers are classified as discrete characters

because both were adopted from Nyffeler and Eggli (1997).

Discrete character states are shown in Table 2, and the

resulting matrix is available in the electronic supplemen-

tary material (Appendix 2).

Molecular dataset

Fresh samples were frozen at -20�C and pulverized to a

fine powder. Before DNA extraction, the tissue powder was

suspended in HEPES buffer (pH 8.0) and centrifuged at

10,000 rpm for 5 min to remove mucilage (Setoguchi et al.

Table 1 continued

Species Acronyms Latitudinal

distribution

Collection data

E. subgibbosa subsp. vallenarensis(F. Ritter) Katt.

E_subgibbosa_val 28� S FK 79 (SGO) (Neotype), Region de Atacama, Maitencillo.

HW 129 (CONC) [JF975699, JF975720], Region de Atacama, west of Vallenar.

E. subgibbosa subsp. wagenknechtii(F. Ritter) Katt.

E_subgibbosa_wag 29� S FK 183 (SGO), Region de Coquimbo, Juan Soldado.

FK 1091 (SGO), Region de Coquimbo, Juan Soldado.

Eggli, Leuenberger and Arroyo-Leuenberger 2878 (CONC), Region de

Coquimbo, Juan Soldado.

HW 44 (CONC) [JF975700, JF975721], Region de Coquimbo, Chungungo.

HW 358, Region de Atacama, Caleta Chanaral.

HW 654, Region de Coquimbo, Cruz Grande.

E. villosa (Monv.) Katt. E_villosa 28�–29� S FK 71 (SGO) (Neotype), Region de Atacama, north of Huasco.

FK 164 (SGO), Region de Atacama, Huayco (=Huasco).

FK 467 (SGO), Region de Atacama, Sarco.

FK 814 (SGO), Region de Atacama, Carrizal Bajo.

FK 1028 (SGO), Region de Atacama, north of Huayco.

Eggli and Leuenberger 2675 (CONC), Region de Coquimbo, Los Choros.

Eggli and Leuenberger 2675 (CONC), Region de Coquimbo, El Trapiche.

Eggli and Leuenberger 2967 (SGO), Region de Coquimbo, Los Choros.

Eggli and Leuenberger 3001 (CONC), Region de Atacama, North of Huayco

(=Huasco).

HW 187 (CONC) [JF975702, JF975725], Region de Atacama, north of Huasco.

HW 477, Region de Atacama, east of Llanos de Challe.

HW 304, Region de Atacama, Aguadas Tongoy.

HW 358, Region de Atacama, Caleta Chanaral.

The collection data included collector and collection number (HW for H.E. Walter and FK for F. Kattermann), region, and locality. Morphological data of the

studied taxa were mainly taken from plants studied in the field (all HW collection numbers). Specimens from herbarium collections are identified in parentheses

(CONC, Universidad de Concepcion; SGO, Museo Nacional de Historia Natural), and when Genbank codes apply they also are given, in square brackets [trnL-trnF,

rpl32-trnL]

Phylogenetics and predictive distribution modeling 117

123

Table 2 Characters and character states for the cladistic analysis of Neoporteria

Character Character states

1. Root system [0] fibrous [1] intermediate [2] taproot

2. Stem elongation [0] columnar

([60 cm)

[1] subcolumnar

(40–60 cm)

[2] globular to

slightly elongate

(5–40 cm)

[3] subglobular

to flattened

(\5 cm)

3. Stem color [0] grey-green to

reddish

[1] grass-green to

yellowish green

4. Spine number [0] 1–20 [1] 21–30 [2] 31–60

5. Central spines thickness [0] thin acicular [1] thick acicular [2] thin hair-like [3] not

applicable

6. Central spines attitude [0] straight to

directed

downwards

[1] straight radiant [2] straight to

recurved

[3] strongly

contorted

[4] not

applicable

7. Radial spines attitude [0] radiant [1] curved upwards [2] pectinate

8. Radial spines thickness [0] majority thin

acicular

[1] majority thick

acicular

[2] thin hair-like

9. External perianth segments attitude [0] straight [1] curved downwards

10. Flower shape [0] wide funnel-

form

[1] funnel-form [2] narrow funnel-

form

[3] tubular

11. Ratio between the pericarpel length

and the hypanthium length

[0] [0.70 [1] 0.50–0.70 [2] 0.35–0.49 [3] \0.35

12. Nectar chamber [0] tubular [1] basally widened,

round

[2] basally

widened, angular

13. Duration of individual anthesis [0] 1–5 days [1] 6–10 days [2] [10 days

14. Total duration of anthesis [0] 3–6 weeks [1] 7–11 weeks [2] 12–24 weeks

15. Main flowering period [0] from end of

winter to spring

[1] spring [2] from autumn to

late winter

16. Flowers and fruits simultaneous [0] yes [1] no

17. Style color [0] whitish to

yellowish

[1] whitish with upper

third reddish

[2] red

18. Wool on tube and pericarpel [0] short and

scant

[1] dense

19. Tube and pericarpel bristles quantity [0] many [1] few [2] absent

20. Fruit shape [0] globular to

subovoid

[1] ovoid [2] clearly

elongated

21. Seed testa surface [0] smooth [1] finely tuberculate [2] ribbed

22. Central spines clearly differentiated

from radials

[0] no [1] yes [2] not applicable

23. Inner perianth segments attitude [0] directed

outwards

[1] intermediate [2] directed

inwards

24. Scotonastic movement [0] absent [1] intermediate [2] present

25. Flower size in relation to spine size [0]

flower \ spines

[1] flower = spine [2]flower [ spines

26. Color of innermost perianth

segments

[0] white or

yellow

[1] bicolorous: inferior

part whitish, superior

fuchsia

[2] fuchsia [3] red

27. Flower color [0] never fuchsia [1] fuchsia

28. Relief of epidermis [0] flat [1] bumpy [2] short-papillate [3] long-

papillate

29. Thickness of epidermis layer

(excluding bulging outer periclinal

walls) (maxima)

[0] 20–30 lm [1] 31–40 lm [2] 41–70 lm

30. Secondary cell divisions of

epidermis cells

[0] Only

periclinal

[1] periclinal or oblique [2] not applicable

118 P. C. Guerrero et al.

123

1998). Total genomic DNA was isolated using the CTAB

procedure described by Doyle and Doyle (1987). Two

genes were amplified using standard primers and proce-

dures (rpl32-trnL(UAG): Shaw et al. 2007; trnL-trnF: Nyff-

eler 2002). When multibands were present in the gel, the

bands corresponding to the size of the gene expected from

the literature were cut out and cleaned using the Wizard�

SV Gel and PCR clean-up system Promega kit (Qiagen).

Fragments were amplified in 25-lL reactions (12.5 lL

GoTaq colorless master (Promega, Madison, USA), 5.0 lL

10 mmol primer, 1.25 lL BSA, 0.0–3.0 lL 10 mmol

MgCl, 4.25–6.25 lL ddH2O) using an automated thermal

cycler and standard PCR procedures, with PCR cycling

time as follows: rpl32-trnL: denaturation at 80�C for 5 min

followed by 45 cycles of denaturation at 95�C for 1 min,

primer annealing at 48–56�C for 1 min, and primer

extension at 65�C for 4 min; followed by a final extension

step of 5 min at 65�C; trnL-trnF: denaturation at 94�C for

4 min, followed by 40 cycles of 94�C for 30 s, 46–54�C for

1 min, and 72�C for 90 s, and finished with a final elon-

gation step at 72�C of 7 min. Sequencing was performed

by Macrogen (Korea).

Sequences were edited using Mega 5.0 and aligned using

Muscle first and then checked manually (Edgar 2004; Tamura

et al. 2007), partitions were concatenated using Mesquite

(Maddison and Maddison 2010) Taxon localities, the asso-

ciated vouchers and the Genbank codes are listed in Table 1.

Alignments are available from the corresponding author.

Phylogenetic analyses were conducted for each indi-

vidual partition (morphology, rpl32-trnL and trnL-trnF)

and for some partition combinations (rpl32-trnL ? trnL-

trnF, rpl32-trnL ? trnL-trnF ? morphology). Bayesian

analyses were implemented using MrBayes version 3.1.2

(Huelsenbeck and Ronquist 2001) The best substitution

model was GTR ? G ? I for each gene and was selected

using the Akaike information criterion in MrModeltest

version 2.0 (Nylander et al. 2004); in the case of mor-

phology we utilized the standard unordered discrete evo-

lution model. The best partitioning strategy between genes

and between genes ? morphology was realized using

comparison of Bayes factors (Nylander et al. 2004), based

on comparing the harmonic mean of the log-likelihoods of

the post-burn-in trees from analyses. These analyses

showed that the best partitioning strategy was the partition

that included both genes (rpl32-trnL ? trnL-trnF) and the

partition that included both genes and morphology (rpl32-

trnL ? trnL-trnF ? morphology). For each separate and

combined analysis, we ran two replicate searches for 5 9

106 generations each, sampling every 1,000 generations

and using default parameters.

In addition, tree searches using maximum likelihood

analysis were conducted to genes partitions using RAxML

version 7.0.3 (Stamatakis et al. 2008). This implementation

of the likelihood method enables optimization of individual

substitution models for different partitions. Thus, we

applied the same combination of models (GTR ? G ? I)

and partitioning strategies between and within genes as in

the Bayesian analyses. Two hundred inferences were exe-

cuted using RAxML on distinct randomized parsimony

starting trees with 1,000 nonparametric bootstrap replicates.

Table 2 continued

Character Character states

31. Number of periclinal and oblique

secondary cell divisions in non-

papillate cells

[0] none [1] few [2] many [3] not

applicable

32. Number of periclinal and oblique

secondary cell divisions in

papillate cells

[0] none [1] few [2] many [3] not

applicable

33. Number of cell layers in the

hypodermis (maxima)

[0] 1–2 [1] 3 [2] 4–7 [3] not

applicable

34. Thickness of the hypodermis layer

(maxima)

[0] 30–50 lm [1] 60–110 lm [2] 140–350 lm [3] not

applicable

35. Firmness of the cortex tissue [0] soft or very

soft

[1] intermediate [2] tough

36. Presence of mucilage in stem

sections

[0] not

mucilagious

[1] slightly or locally

mucilaginous

[2] distinctly

mucilaginous

[3] intensively

mucilaginous

37. Color of the central and inner cortex [0] pale greenish

or whitish

[1] intermediate [2] green [3] reddish

38. Ratio of pith to plant diameter (in

transverse sections at the widest

diameter)

[0] 0.15–0.20 [1] 0.22–0.28 [2] 0.30–0.45

Characters 28 to 38 and their states (apart from character state 37-3) were adopted from Nyffeler and Eggli (1997)

Phylogenetics and predictive distribution modeling 119

123

Predictive distribution modeling and the geography

of divergence

We used predictive distribution modeling (also known as

climatic niche modeling) to infer taxa range extent asso-

ciated to the spatial distribution of climatic suitability. We

modeled the distribution of sister taxa using Maxent ver-

sion 3.3.3e (Phillips et al. 2004, 2006). Two types of data

are required to predict species ranges: environmental data

and information on their known occurrence. The environ-

mental layers consisted of the Bioclim climatic datasets

which summarize temperature and precipitation dimen-

sions in the environment (Hijmans et al. 2005). In spite of

some criticisms (heterogeneous distribution of meteoro-

logical stations supplying data to construct the interpolated

climatic models), Bioclim is the most complete climatic

database available for Chile. With 19 layers and the

appropriate formats for Geographic Information Systems,

Bioclim has become widely used in evolutionary and bio-

diversity studies (Kozak and Wiens 2006; Raxworthy et al.

2007; Evans et al. 2009; Zizka et al. 2009).

Information on locality data was obtained from different

sources: field excursions, literature and Chilean herbaria

(CONC = Universidad de Concepcion; SGO = Museo

Nacional de Historia Natural). Cacti are not as well repre-

sented in herbaria as other families because of difficulties

arising from drying and storing their tissues, but they have a

long history of botanic and amateur work reflected in liter-

ature and in informal sources. To validate literature and

herbaria localities, we verified the majority of the known

populations in the field (Fig. 1). In this study, we included

unpublished and new localities for most of the Chilean taxa.

The number of localities per taxon ranged from four

(E. senilis subsp. coimasensis) to 46 (E. subgibbosa var.

subgibbosa). Maxent is considered the best model for man-

aging datasets of small sample sizes, and its accuracy is

proven to be greater for species with small geographic ranges

and limited environmental tolerance as is the case in most

Neoporteria taxa (Hernandez et al. 2006; Wisz et al. 2008).

Using Maxent, we calculated the average niche-based

occurrence probabilities of 10 replicated models and we

mapped probabilities [0.50 and [0.75. Overlapping was

assessed by analyzing the probability of the co-occurrence

of a pair of sister taxa obtained by multiplying their

occurrence probabilities across their predicted distributions

(for each pixel of 1 km2 of the grids). The overlap per-

centage between both species was calculated by obtaining

the area in which the probability of co-occurrence

was [0.75 compared to the total range extent. Overlaps

may vary between 0% (completely allopatric) and 100%

(completely sympatric).

After running preliminary phylogenetic analyses we

decided to add spatial evaluation of E. chilensis, since its

floral morphology suggests an intermediate position

between subsections Neoporteria and Horridocactus. As

E. chilensis is restricted to a small area in which two

Neoporteria taxa (E. subgibbosa var. subgibbosa and var.

litoralis) and a species from subsection Horridocactus

(E. curvispina (Bertero ex Colla) Katt. var. mutabilis

(F.Ritter) Katt.) grow sympatrically, we checked if the

overlap between the former species and each of the Neo-

porteria taxa matched the distribution range of E. chilensis.

Results

Phylogenetic relationships

The morphological data placed E. islayensis and E. aurata

in a basal polytomy together with Corryocatus brevistylus

(Fig. 2). All the Neoporteria sensu stricto taxa were

included in a single well-supported clade that also com-

prised E. taltalensis (subsection Horridocactus) and

E. chilensis in a basal position. The subclade comprising

E. odieri and E. krausii, together with the rest of the sub-

species of the E. taltalensis complex also received strong

support. The sister pairs receiving strong support were

E. taltalensis/E. sociabilis, E. senilis subsp. senilis/subsp.

coimasensis and E. subgibbosa subsp. clavata/subsp. nig-

rihorrida, whereas the pair E. subgibbosa subsp. subgibb-

osa/subsp. nigrihorida received less support.

Bayesian and maximum likelihood analyses using single

molecular markers revealed similar general phylogenetic

relationships (Fig. 2), the basal species remained in the

same position; in the rpl32-trnL partition that included

Neowerdermannia chilensis, the latter was placed between

E. aurata and E. curvispina subsp. curvispina. The most

important difference between the trnL-trnF and rpl32-

trnL analyses was the placement of E. sociabilis and

E. taltalensis var. taltalensis: in the former analysis

E. taltalensis and E. sociabilis were placed within the

Neoporteria clade, whereas the latter placed it in the sister

clade to Neoporteria sensu stricto compromising E. odieri,

E. krausii, E. paucicostata, and the other taxa of the

E. taltalensis complex. This difference might be explained

by the fact that the rpl32-trnL partition included more taxa

compared with the trnL-trnF partition.

The topologies of the trees resulting from analyses that

combined all molecular data and molecular data plus mor-

phology were almost identical, only some node supports

changed. In both partitions E. sociabilis and E. taltalensis

var. taltalensis were excluded from the Neoporteria clade

(Fig. 3). Sister pairs that were strongly supported were

E. senilis subsp. senilis/subsp. coimasensis, E. subgibbosa

subsp. clavata/subsp. nigrihorrida, E. subgibbosa subsp.

subgibbosa/subsp. nigrihorrida. As relationships among

120 P. C. Guerrero et al.

123

E. villosa, E. subgibbosa subsp. vallenarensis/ wagen-

knechtii were poorly resolved in most analyses, we studied

their distribution overlap.

Predictive distribution modeling and the geography

of divergence

Predictive distribution modeling of species showed that the

degree of geographic overlap between sister taxa is null

(Fig. 4). All taxa showed \0.5 probabilities of co-occur-

rence between them, indicating an extensive allopatric

distribution of relative taxa. Eriosyce senilis showed three

disjunctions, one corresponding to subspecies coimasensis,

the other two corresponding to the predicted areas of

occurrence of E. senilis subsp. senilis that are separated by

a large unsuitable intervening region (Fig. 4a). The sister

pair E. subgibbosa subsp. clavata/E. subgibbosa subsp.

nigrihorrida are distributed allopatrically, although the

core area of their distributions (P [ 0.75) is separated by a

stretch of only 5 km (Fig. 4b). The group comprising

E. subgibbosa subsp. vallenarensis, subsp. wagenknechtii

and E. villosa showed a similar allopatric pattern (occur-

rence probabilities \0.5) (Fig. 4c). E. subgibbosa var.

subgibbosa and var. castanea were shown to be separated

by a distance of some 20 km (Fig. 4d).

In addition to that, predictive distribution modeling

indicated that E. subgibbosa var. litoralis and E. curvipina

var. mutabilis have null overlap between them, while the

probability of an overlap of the latter species’ distribution

with E. subgibbosa var. subgibbosa was high (Fig. 5a). The

Fig. 1 Population localities of

Eriosyce subgen. Neoporteriacompiled from field excursions,

literature and Chilean herbaria

Phylogenetics and predictive distribution modeling 121

123

overlap between these species matched 77.2% of the dis-

tribution extent of E. chilensis (P [ 0.75; Fig. 5b).

Discussion

Phylogenetic relationships

The monophyly of Neoporteria sensu stricto has never

been questioned seriously, as reflected by the fact that it

was always seen as a distinct group, although at different

taxonomic ranks (genus, subsection, subgroup, subgenus;

supplementary material). When Kattermann (1994), how-

ever, broadened the concept of Neoporteria sensu stricto,

proposing to place E. chilensis within subsection Neopor-

teria, some dispute arose, as the presentation of the case

was contradictive: the cladogram (op. cit.) resulting from

the phylogenetic analysis of Eriosyce sensu lato, did not

place E. chilensis within the monophyletic subsection

Neoporteria, but with subsection Horridocactus (Walter

Fig. 2 Phylogeny of Eriosyce subgen. Neoporteria based on separate

Bayesian analysis of morphology, and separate Bayesian and

maximum likelihood analyses of molecular data based on trnL-

trnF and rpl32-trnL (chloroplast DNA). Numbers above branches

indicate a posteriori Bayesian support; numbers below branchesindicate ML bootstrap support. Filled black circles indicate nodes that

were supported with [50% of bootstrap support in ML analyses

122 P. C. Guerrero et al.

123

2008). E. chilensis, an insect-pollinated species with wide

funnel-form flowers, had before been considered a member

of Pyrrhocactus (Ritter 1980; Hoffmann 1989; Zuloaga

et al. 2008) or subgenus Horridocactus (Hoffmann and

Walter 2004) for this reason. Kattermann (1994), however,

placed it within subsection Neoporteria, arguing that the

wide funnel-form flowers might indicate a reversal from

hummingbird pollination to bee pollination. Nyffeler and

Eggli’s (1997) cladogram based on stem morphology and

anatomical data suggested that E. chilensis is the most

basal member of subsection Neoporteria.

The historical taxonomic complexity of the placement

of E. chilensis either within Horridocactus /Pyrrhocactus

or Neoporteria (Appendix 1) was reflected by our phylo-

genetic analyses, as different partitions of our dataset

indicated two different positions of E. chilensis within the

Neoporteria clade: in the analyses of combined molecular

datasets E. chilensis was placed in the unresolved terminal

subclade, whereas in the morphology-based tree it was

placed basal to all ingroup taxa, most probably because of

the various ‘‘pollination syndrome characters’’ in our

morphological data set. This assumption coincides with

Styles (1981), who reported that nearly all hummingbird

flowers of North America have evolved from insect-polli-

nated species. On the other hand, in the analysis of Nyffeler

and Eggli (1997), whose character set did not comprise any

floral characters, it was also placed in a basal position

within Neoporteria.

These contradictive results suggest that the morphology

of E. chilensis may comprise several homoplasies, making

it difficult to decide its phylogenetic placement. One

plausible explanation for these homoplasies is that E. chil-

ensis might share morphological traits between subsections

Neoporteria and Horridocactus. Predictive distribution

models showing the overlap of the possible parental spe-

cies suggest that hybridization is plausible (see below).

Kattermann’s (1994) broad concept of E. subgibbosa—

including five heterotypic subspecies—was not corrobo-

rated by our analyses (morphological and molecular),

because E. subgibbosa subsp. subgibbosa was never

grouped in one clade. Our analysis based on the combined

molecular markers (Fig. 3a) suggests the existence of at

least three distinct entities within Kattermann’s (1994)

broad concept of E. subgibbosa:

1. E. subgibbosa var. subgibbosa/var castanea;

2. E. subgibbosa subsp. nigrihorrida/subsp. clavata; and

3. E. subgibbosa subsp. wagenknechtii/subsp. vallenar-

ensis together with E. villosa.

This supports Walter’s (2008) assumption that E. sub-

gibbosa subsp. wagenknechtii/subsp. vallenarensis are

much closer related to E. villosa than to E. subgibbosa.

Four different taxa were placed in the unresolved ter-

minal subclade—E. chilensis, E. senilis, and E. subgibbosa

subsp. nigrihorrida, subsp. clavata and var. litoralis—but,

unexpectedly, our molecular data were not able to resolve

the relationships between them. The exclusion of E. sub-

gibbosa subsp. clavata, subsp. nigrihorrida and var. lito-

ralis from the E. subgibbosa subsp. subgibbosa subclade,

however, indicates that they might represent different

species, a finding corroborating their classification by

Walter (2008). A future study would probably have to

Fig. 3 Phylogeny of Eriosyce subgen. Neoporteria based on com-

bined Bayesian and maximum likelihood analyses of molecular data:

trnL-trnF ? rpl32-trnL (chloroplast DNA), and combined Bayesian

analysis of molecular and morphological data. Numbers above

branches indicate a posteriori Bayesian support; numbers belowbranches indicate ML bootstrap support. Filled black circles indicate

nodes that were supported with [50% of bootstrap support in ML

analyses

Phylogenetics and predictive distribution modeling 123

123

124 P. C. Guerrero et al.

123

sample inter and infra-populational variability to resolve

relationships in this subclade.

The analyses based on the combined molecular (Fig. 3a)

and molecular/morphological data (Fig. 3b) excluded

E. sociabilis and E. taltalensis subsp. taltalensis from the

Neoporteria clade, whereas our morphological and trnL-

trnF (Fig. 2a, b) data included it in this clade. The cause of

this inconsistency might be the less dense taxon sampling

in the trnL-trnF partition compared with the rpl32-

trnL partition. The gross morphologies of E. sociabilis and

E. taltalensis subsp. taltalensis are strikingly similar (Ritter

1980; Walter 2008) and they share many flower characters,

for example the fuchsia-red color, and the size or the

narrow funnel-form hypanthia; the inner perianth segments

of E. sociabilis are, however, erect to sometimes somewhat

inclined inwards, one of the typical characters of a hum-

mingbird syndrome. For these reasons, E. sociabilis has

been considered a member of Neoporteria sensu stricto (a

hummingbird-pollinated group) by all authors since Ritter

(1963) described it as Neoporteria sociabilis. Our molec-

ular data, however, placed E. sociabilis within the highly

supported sister clade to Neoporteria comprising (amongst

other taxa) the taxa of the E. taltalensis complex, thus

suggesting that in the circumscription of Kattermann

(1994), subsection Neoporteria is paraphyletic.

After Schlumpberger and Raguso (2008) had shown for

a group of closely related species of Echinopsis that floral

syndromes are particularly unreliable (Nyffeler and Eggli

2010), the exclusion of E. sociabilis from the Neoporteria

clade by our molecular data seems to present another case

documenting the fact that molecular analyses (Ritz et al.

2007) provide ample clarity that pollination syndromes are

highly plastic and that there is convincing evidence that the

same floral syndrome can evolve in parallel in the same

clade (Nyffeler and Eggli 2010). It is noteworthy to men-

tion that not only floral syndromes but also hummingbirds

must be considered as ‘‘unreliable’’, as they are not as

Fig. 5 Predicted geographic distributions of two subsections within

Eriosyce whose distributions overlap in central coastal Chile. Circlesare documented localities for each taxon. We evaluated the proba-

bility of co-occurrence of one Horridocactus and two Neoporteriataxa to determine if the estimated overlaps matched E. chilensisdistribution. a Geographic distributions of E. subgibbosa var.

litoralis/var. subgibbosa (subsect. Neoporteria) and E. curvipinavar. mutabilis (subsect. Horridocactus). b overlap between E.subgibbosa var. subgibbosa and E. curvispina var. mutabilis and

distribution of E. chilensis. We did not detect overlap with E.subgibbosa var. litoralis

Fig. 4 Predicted geographic distributions of selected sister taxa of

Eriosyce subgen. Neoporteria based on ecological niche modeling.

Circles are documented localities for each taxon. a E. senilis subsp.

senilis and subsp. coimasensis. b E. subgibbosa subsp. nigrihorridaand E. subgibbosa subsp. clavata. c E. subgibbosa subsp. vallenar-ensis, subsp. wagenknechtii and E. villosa. d E. subgibbosa var.

subgibbosa and var. castanea

b

Phylogenetics and predictive distribution modeling 125

123

specialized on ‘‘hummingbird flowers’’ as often suggested:

they also might feed on non-tubular flowers with wide open

corollas, for example those in E. taltalensis and E. chil-

ensis—especially during the winter when, e.g., winter-

blooming E. taltalensis subsp. taltalensis flowers are often

the only flowers available (Walter 2008).

Predictive distribution modeling and the geography

of divergence

Analyses of the degree of geographic overlap between sister

taxa obtained from the cladistic analysis showed that most

of the sister taxa are spatially segregated entities or have

minimum overlap, suggesting that allopatric divergence is a

widespread phenomenon in Neoporteria (Fig. 4). None of

the sister taxa, (E. subgibbosa subsp. clavata / subsp. nig-

rihorrida; E. subgibbosa var. subgibbosa/ var. castanea;

E. subgibbosa subsp. wagenknechtii/subsp. vallenarensis/

E. villosa; E. senilis subsp. senilis/ subsp. coimasensis) had

any degree of predictable geographic overlap (Fig. 4).

Although it was not part of the objectives of this study to

distinguish between the different types of allopatry, some

particular aspects may be considered. Asymmetric range

sizes between sister taxa have often been associated with

peripatric speciation (Losos and Glor 2003); however, our

sister taxa have similar range sizes (Fig. 4). This strongly

suggests that their divergence originated from an allopatric

vicariant event (probably because of climatic changes) and

the low overlap may have been caused by recent and subtle

changes in their distribution ranges (caused by climatic

changes). Moreover, these taxa are located in the transi-

tional zone between summer and winter rainfall, and the

transitional zone between the hyper-arid and semi-arid

ombrotypes of northern-central Chile which underwent

continuous changes during the Holocene and Pleistocene

(Latorre et al. 2002; Houston 2006; Luebert and Pliscoff

2006). During the Pleistocene, periods of glaciations cre-

ated a series of wet–dry cycles, enabling the expansion of

the range of the northernmost Neoporteria taxa during wet

periods (Latorre et al. 2002; Maldonado et al. 2005), fol-

lowed by a subsequent contraction of ranges, thus pro-

moting population isolation during cycles of dry periods.

During the Holocene, the climate became more arid

(Maldonado et al. 2005), restricting the distribution of the

northern Neoporteria taxa. Indeed, some of these popula-

tions might have survived in zones with oceanic fog

dynamics (Cereceda et al. 2002, 2008), thus favoring long-

term divergence between them. Other vicariant relict spe-

cies from various plant families, for example Griselinia

G.Forst. (Griseliniaceae), Heliotropium section Cochranea

(Miers) Kuntze (Heliotropiaceae), Moscharia Ruiz et Pav.

(Asteraceae) and Tillandsia L. (Bromeliaceae), are pre-

served in the northern fog oasis system. Although their

separation had occurred in different geological eras, aridity

had always contributed to their allopatric divergence

(Dillon and Munoz-Schick 1993; Katinas and Crisci 2000;

Luebert and Wen 2008; Zizka et al. 2009; Guerrero et al.

2011).

The predicted spatial separation between E. senilis

subsp. senilis / E. senilis subsp. coimasensis is approxi-

mately 50–100 km, suggesting that both taxa may belong

to different evolutionary lineages and may even represent

distinct species (Wiens 2004a). Interestingly, two spatially

separated entities in E. senilis subsp. senilis, can be clearly

distinguished from their predicted distributions: the popu-

lations between latitudes 30�–31� S (the Elqui and Limarı

Valleys and adjacent regions) and the populations between

31�300–32� S (Choapa Valley and adjacent regions). The

extensive unsuitable intervening regions between these

areas make gene flow between these populations unlikely.

Yet, on the basis of their morphological similarity, they

were seen as one taxonomic entity (i.e. E. senilis subsp.

senilis) by Hoffmann and Walter (2004) and Hunt et al.

(2006). In the past, some authors had considered the two

populations as distinct taxa, classifying the northern pop-

ulations as E. senilis subsp. elquiensis Katt., respectively

Neoporteria nidus var. gerocephala (Y. Ito) Ritt., and the

southern populations as E. senilis subsp. senilis, corre-

spondingly N. multicolor F. Ritter (Ritter 1980; Katter-

mann 1994). Our spatial analysis supports Kattermann’s

and Ritter’s division of this taxon into two taxonomic units.

Less distance separates E. subgibbosa var. subgibbosa

from E. subgibbosa var. castanea (*20 km) together with

little morphological difference, suggesting that climatic

barriers and the resulting reproductive isolation were

imposed more recently and thus the divergence between

the two taxa is still in an early stage.

Predictive distribution models showed that the overlap

between E. subgibbosa var. subgibbosa and E. curvispina

var. mutabilis matches with a high percentage of the dis-

tribution of E. chilensis, suggesting that both might be

parental species of E. chilensis. However, phylogenetic

analyses placed E. chilensis with E. subgibbosa var. lito-

ralis suggesting genetic proximity and a complex scenario

of species reproductive interactions. Further study is nee-

ded to test E. chilensis’ reticular origin, and extensive

genetic sampling at the population level and detailed

evaluations of the reproductive biology of the four species.

Predictive distribution modeling and the phylogenetic

analyses indicate that most of our taxonomic units corre-

spond to divergent evolutionary lineages with different

temporal stages of speciation and suggest that these lin-

eages mostly originated as a result of historical vicariant

events that resulted in the splitting of one common ancestor

into two new taxa. The role of pollinators in driving spe-

ciation seems to be less important than geographic isolation

126 P. C. Guerrero et al.

123

caused by climatic tolerances, although, less frequently,

pollinator guilds and the phenology of species might be

important in reticular speciation.

Acknowledgments We are grateful to A. Marticorena (CONC),

M. Munoz-Schick (SGO), and I. Meza (SGO) for access to their

collection of specimens. We appreciate R. Scherson’s advice on DNA

extraction and amplification. We are indebted with P. Posadas,

J.V. Crisci, U. Eggli and R. Nyffeler, for their valuable comments and

collaboration on different stages of this study. We are also grateful to

CONAF for access to the National System of Protected Areas.

Computational resources were provided by the Bioportal at the

University of Oslo, Norway (http://www.bioportal.uio.no). This

research was partially supported by the Instituto de Ecologıa y Bio-

diversidad, project ICM-P05-002. PCG acknowledges CONICYT

doctoral fellowship and thesis support.

References

Anderson EF (2001) The cactus family. Timber Press, Portland

Barraclough TG, Vogler AP (2000) Detecting the geographical

pattern of speciation from species-level phylogenies. Am Nat

155:419–433

Cereceda P, Osses P, Larraın H, Farıas M, Lagos M, Pinto R,

Schemenauer RS (2002) Advective orographic and radiation fog

in the Tarapaca region Chile. Atmos Res 64:261–271

Cereceda P, Larraın H, Osses P, Farıas M, Egana I (2008) The spatial

and temporal variability of fog and its relation to fog oasis in the

Atacama Desert Chile. Atmos Res 87:312–323

Darwin C (1859) The origin of species by means of natural selection

or the preservation of favoured races in the struggle for life. John

Murray, London

Dillon MO, Munoz-Schick M (1993) A revision of the dioecious

genus Griselinia (Griseliniaceae) including a new species from

the coastal Atacama Desert of northern Chile. Brittonia

45:261–274

Doyle JJ, Doyle JA (1987) A rapid DNA isolation procedure for small

quantities of fresh leaf tissue. Phytochem Bull 19:11–15

Edgar RC (2004) MUSCLE: multiple sequence alignment with high

accuracy and high throughput. Nucleic Acids Res 32:1792–1797

Evans MEK, Smith SA, Flynn RS, Donoghue MJ (2009) Climate

niche evolution and diversification of the ‘‘bird-cage’’ evening

primroses (Oenothera sections Anogra and Kleinia). Amer

Naturalist 173:225–240

Ferryman R (2003) Notulae Systematicae Lexicon Cactacearum

Spectantes IV. In: Hunt D, Taylor N (eds.) Cactaceae Systematic

Initiatives 16:11

Fitzpatrick BM, Turelli M (2006) The geography of mammalian

speciation: mixed signals from phylogenies and range maps.

Evolution 60:601–615

Futuyma DJ (2005) Evolutionary biology. Sinauer Associates,

Sunderland

Graham CH, Ron RR, Santos JC, Schneider CJ, Moritz C (2004)

Intergrating phylogenetics and environmental niche models to

explore speciation mechanisms in Dendrobatid frogs. Evolution

58:1781–1793

Guerrero PC, Duran AP, Walter HE (2011) Latitudinal and altitudinal

patterns of the endemic cacti from the Atacama Desert to Mediter-

ranean Chile. J Arid Environ. doi:10.1016/j.jaridenv.2011.04.036

Hernandez PA, Graham CH, Master LL, Albert DL (2006) The effect

of sample size and species characteristics on performance of

different species distribution modeling methods. Ecography

29:773–785

Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very

high resolution interpolated climate surfaces for global land

areas. Int J Climatol 25:1965–1978

Hoffmann AE (1989) Cactaceas en la flora silvestre de Chile.

Fundacion Claudio Gay, Santiago de Chile

Hoffmann AE, Walter HE (2004) Cactaceas en la flora silvestre de

Chile. Fundacion Claudio Gay, Santiago de Chile

Houston J (2006) Variability of precipitation in the Atacama Desert:

its causes and hydrological impact. Int J Climatol 26:2181–2198

Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference

of phylogenetic trees. Bioinformatics 17:745–755

Hunt DR, Taylor N, Charles G (2006) The New Cactus Lexicon. DH

books, Milborne Port

Hunt DR (2003) In: Hunt and Taylor (eds) Notulae Systematicae Lexicon

Cactacearum Spectantes IV. Cactaceae Systematic Initiatives 16:9

Katinas L, Crisci JV (2000) Cladistic and biogeographic analyses of

the genera Moscharia and Polyachyrus (Asteraceae Mutisieae).

Syst Bot 25:33–46

Katinas L, Crisci JV (2008) Reconstructing the biogeographical

history of two plant genera with different dispersion capabilities.

J Biogeogr 35:1374–1384

Kattermann F (1994) Eriosyce (Cactaceae) the genus revised and

amplified. Succul Plant Res 1:1–176

Kattermann F (2001) Nomenclatural adjustments in Eriosyce Phil.

Cactaceae Syst Ini 12:14

Kozak KH, Wiens JJ (2006) Does niche conservatism promote

speciation? A case study in North American salamanders.

Evolution 60:2604–2621

Latorre C, Betancourt JL, Quade J (2002) Vegetation invasions into

absolute desert: A 45000 yr rodent midden record from the

Calama-Salar de Atacama basins northern Chile (lat 22�–24�S).

Geol Soc Am Bull 114:349–366

Losos JB, Glor RE (2003) Phylogenetic comparative methods and the

geography of spciation. Trends Ecol Evol 18:220–227

Luebert F, Pliscoff P (2006) Sinopsis bioclimatica y vegetacional de

Chile. Editorial Universitaria, Santiago de Chile

Luebert F, Wen J (2008) Phylogenetic analysis and evolutionary

diversification of Heliotropium Sect Cochranea (Heliotropia-

ceae) in the Atacama Desert. Syst Bot 33:390–402

Maddison WP, Maddison DR (2010) Mesquite: a modular system for

evolutionary analysis. Version 2.73 http://mesquiteproject.org

Mayr E (1959) Isolation as an evolutionary factor. Proc Natl Acad Sci

USA 103:221–230

Maldonado A, Betancourt JL, Latorre C, Villagran C (2005) Pollen

analyses from a 50000-yr rodent midden series in the southern

Atacama Desert (25� 300S). J Quat Sci 20:493–507

Nixon KC, Davis JI (1991) Polymorphic taxa missing value and

cladistic analysis. Cladistics 7:233–241

Nyffeler R, Eggli U (1997) Comparative stem anatomy and system-

atics of Eriosyce sensu lato (Cactaceae). Ann Bot 80:767–786

Nyffeler R (2002) Phylogenetic relationships in the cactus family

(Cactaceae) based on evidence from trnK/matK and trnL-trnF

sequences. Am J Bot 89:312–326

Nyffeler R, Eggli U (2010) A farewell to dated ideas and concepts—

molecular phylogenetics and a revised suprageneric classifica-

tion of the family Cactaceae. Schumannia 6:109–149

Nylander JA, Ronquist AF, Huelsenbeck JP, Nieves-Aldrey JL (2004)

Bayesian phylogenetic analysis of combined data. Syst Biol

53:47–67

Phillips SJ, Dudık M, Schapire RE (2004) A maximum entropy

approach to species distribution modeling In: Proceedings of the

twenty-first international conference on machine learning. ACM

Press, New York

Phylogenetics and predictive distribution modeling 127

123

Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy

modeling of species geographic distributions. Ecol Model

190:231–259

Posadas P, Crisci JV, Katinas L (2006) Historical biogeography: a

review of its basic concepts and critical issues. J Arid Environ

66:389–403

Raxworthy CJ, Ingram CM, Rabibisoa N, Pearson RG (2007)

Applications of ecological niche modeling for species delimita-

tion: a review and empirical evaluation using day Geckos

(Phelsuma) from Madagascar. Syst Biol 56:907–923

Ritter F (1963) Succulenta (NL) 1:3

Ritter F (1980) Kakteen in Sudamerika vol 3. Selbstverlag,

Spangenberg

Ritz CM, Martins L, Mecklenburg R, Goremykin V, Hellwig FH

(2007) The molecular phylogeny of Rebutia (Cactaceae) and its

allies demonstrates the influence of paleogeography on the

evolution of South American mountain cacti. Am J Bot

94:1321–1332

Savolainen V, Anstett MC, Lexer C, Hutton I, Clarkson JJ, Norup

MV, Powell MP, Springate D, Salamin N, Baker WJ (2006)

Sympatric speciation in palms on an oceanic island. Nature

441:210–213

Schlumpberger BO, Raguso RA (2008) Geographic variation in floral

scent of Echinopsis ancistrophora (Cactaceae); evidence for

constraints on hawkmoth attraction. Oikos 117:801–814

Schemske DW, Bradshaw HD (1999) Pollinator preference and the

evolution of floral traits in monkeyflowers (Mimulus). Proc Natl

Acad Sci USA 96:11910–11915

Setoguchi H, Osawa TA, Pintaud JC, Jaffre T, Veillon JM (1998)

Phylogenetic relationships within Araucariaceae based on rbcL

genes sequences. Am J Bot 85:1507–1516

Shaw J, Lickey EB, Schilling EE, Small RL (2007) Comparison of

whole chloroplast genome sequences to choose noncoding

regions for phylogenetic studies in angiosperms: the tortoise

and the hare III. Am J Bot 94:275–288

Stamatakis A, Hoover P, Rougemont J (2008) A rapid bootstrap

algorithm for the RAxML web-servers. Syst Biol 75:758–771

Styles EG (1981) Geographical aspects of bird-flower coevolution,

with particular reference to Central America. Ann Missouri Bot

Gard 68:323–351

Tamura K, Dudley J, Nei M, Kumar S (2007) Molecular evolutionary

genetics analysis (MEGA) software version 4.0. Mol Biol Evol

24:1596–1599

Turelli M, Barton NH, Coyne JA (2001) Theory and speciation.

Trends Ecol Evol 16:330–3438

Walter HE, Machler W (2006) Rearrangements in the systematics of

Eriosyce napina (Philippi) Kattermann (Cactaceae). Cactus

World 24(3):135–143

Walter HE (2008) Floral biology phytogeography and systematics of

Eriosyce subgenus Neoporteria (Cactaceae). Bradleya 26:75–98

Wiens JJ (2004a) What is speciation and how should we study it? Am

Nat 163:914–923

Wiens JJ (2004b) Speciation and ecology revisited: phylogenetic

niche conservatism and the origin of species. Evolution

58:193–197

Wiens JJ, Graham CH (2005) Niche conservatism: integrating

evolution ecology and conservation biology. Annu Rev Ecol

Syst 36:519–539

Wisz MS, Hijmans RJ, Li J, Peterson AT, Graham CH, Guisan A

(2008) Effects of sample size on the performance of species

distribution models. Diversity Distrib 14:763–773

Zizka G, Schmidt M, Schulte K, Novoa P, Pinto R, Konig K (2009)

Chilean Bromeliaceae: diversity, distribution and evaluation of

conservatio status. Biodiversity Conserv 18:2449–2471

Zuloaga FO, Morrone O, Belgrano MJ (2008) Catalogo de plantas

vasculares del cono sur (Argentina, sur de Brasil, Chile,

Paraguay y Uruguay). Instituto de Botanica Darwinion. Avail-

able via http://wwwdarwineduar/Proyectos/FloraArgentina/

FAasp. Accessed 28 May 2009

128 P. C. Guerrero et al.

123